Research And Simulated Implementation Of Fair Intelligent QoS Routing Mechanism In NGI

It has been the urgent demand to enjoy the high quality network service any time and any place in NGI (Next Generation Internet). Obviously, ABC (Always Best Connected)-supported at the beginning of and during the communication is the Internet development trend. That is, allow users both to connect and use services provided by NGI with the best way, support users to adjust end-to-end routing transparently and adaptively any time and any place among communication, and support QoS (Quality of Service) seamless roaming.In this thesis, the research mainly focuses on the ABC-supported initial routing algorithms in NGI, including unicast and multicast routing. To better support ABC, there are many aspects concerned in routing, such as user QoS requirements and satisfaction degree, user preference, user desire charge to pay for applications, profit sum of ISPs (Internet Service Provider), profit of each ISP, reasonable assignment of network resource, and network load balance, etc. To implement "all-win" in deed, a combinative mechanism should be proposed, including Nash-based and Pareto optimum-based microeconomic theory and method, fuzzy mathematics and non-cooperative pure-strategy-based two-player game. To show the fairness between user and ISP, a simple resource management and assignment is proposed, plusing a dynamic pricing strategy. Furthermore, we consider charges apportion fairness among multicast members, and profits apportion fairness among ISPs, etc. On the NP-complete problem, two swarm intelligent algorithms are proposed based on AntNet algorithm and BeeHive algorithm to find QoS-constrainted paths or multicast trees, respectively. We hope that our research can speed up the realization of NGI.To verify the feasibility and evaluate the performance of the proposed two algorithms, a simulated model is created and encoded in NS2 simulating platform. And then, the results of the simulation are analyzed. By simulation research, satisfactory results are obtained. It is shown that the proposed two algorithms can solve the problem of ABC-supported fair intelligent QoS routing in NGI efficiently.